Patent Grant
7508853
Regenerative amplifiers utilizing chirped pulse amplification (CPA) have been the dominant means for obtaining pulse energies greater than a microjoule with pulse durations in the femtosecond to picosecond range. Microjoule to millijoule pulse energies with pulse durations below 10 picoseconds have been found to be particularly useful for micromachining and for medical applications such as Lasik. However, a big stumbling block in the utilization of ultrafast sources for these applications has been that the regenerative amplifier is more of a piece of laboratory equipment and not conducive to the industrial setting.
Alternative sources for microjoule level, ultrafast pulses are emerging; utilizing all fiber chirped pulse amplification designs. Such systems are inherently more stable since they are based on technology similar to that utilized in Telecomm systems. During the past decade, there has been intensive work and success in making such systems practical. However, for higher pulse energies in the millijoule range, regenerative amplifiers will continue to dominate for some time since pulse energies above a millijoule have not been demonstrated in an all fiber system.
For micromachining applications, more industrially compatible regenerative amplifiers are now being developed based on Nd: or Yb: doped materials, rather than the Ti:sapphire that has dominated the scientific market. There are two basic reasons for this change. Commercial markets typically do not require the shorter pulses that can only be obtained from the Ti:sapphire regenerative amplifier, and the Nd: and Yb: based materials can be directly diode pumped, which makes these systems more robust and less expensive. An unresolved technical issue for Nd: or Yb: based regenerative amplifiers is the need for an equally robust seed source for femtosecond or picosecond pulses. The present seed lasers are mode-locked solid-state lasers with questionable reliability. It would be preferable to have a robust fiber seed source similar to that which has been developed for the Ti:sapphire regenerative amplifier, and used where Ti:sapphire regenerative amplifiers are applied to more commercial applications.
In a copending U.S. application Ser. No. 10/960,923, filed which is assigned to the common assignee and the disclosure of which is incorporated by reference in its entirety, the design changes needed for a mode-locked Yb:doped fiber oscillator and amplifier to be utilized as a seed source for a Yb: or Nd: based solid-state regenerative amplifier are described. The purpose of this application is to modify and apply many of the improvements in all fiber chirped pulse amplification systems for application to the seed source of a regenerative amplifier.
The purpose of this invention is to incorporate the many recent improvements in femtosecond mode-locked fiber lasers and femtosecond fiber chirped pulse amplification systems to regenerative amplifier systems that incorporate femtosecond or picosecond pulse sources based on fiber seed-sources and/or fiber amplifiers.
Yb: and Nd: mode-locked oscillators with fiber amplifiers can be utilized as sources of ultrafast pulses for regenerative amplifiers in order to obtain higher pulse energies than can be realized at this time from all fiber short pulse systems. Ser. No. 10/960,923 (incorporated by reference herein) describes specifically how the sources can be configured to be implemented in such fiber based seed sources for solid state regenerative amplifiers. The femtosecond source and fiber amplifier need to be carefully configured in order to obtain optimum, reliable performance when incorporated into such a system. Recently there have been many improvements in mode-locked fiber sources implemented with fiber amplifiers in chirped pulse amplifier systems that can be utilized in a regenerative amplifier system that typically is based on chirped pulse amplification. Applicable improvements to fiber mode-locked sources are disclosed in Ser. Nos. 09/576,772, 09/809,248, 10/627,069, 10/814,502 and 10/814,319 (all incorporated by reference herein). Alternative suitable femtosecond sources that utilize fiber amplification for pulse conditioning and shortening are described in Ser. No. 10/437,057. One of the difficulties with chirped pulse amplification systems has been in producing reliable and compact pulse stretchers that can be dispersion matched to pulse compressors suitable for high pulse energies.
Significant improvements for dispersion matched fiber stretchers for fiber based chirped pulse amplification are disclosed in Ser. No. 10/992,762, filed Nov. 22, 2004 (incorporated by reference herein). These improvements are also applicable to chirped pulse amplification systems even when solid state bulk mode-locked lasers are utilized as the seed source. Significant improvements have been made in packaging, electronic controls, fabrication processes and optical parameter controls in order to make fiber based femtosecond sources reliable. These engineering improvements can also be utilized in these regenerative amplifier systems and are disclosed in Ser. Nos. 10/606,829, 10/813,163, 10/813,173 and 11/024,948 (all incorporated by reference herein).
Previously, Yb: and Nd: mode-locked oscillators and fiber amplifiers have been utilized as pulse sources for narrow bandwidth, bulk, solid-state amplifiers including regenerative amplifiers that can produce pulses 20 picoseconds or greater. In general, the configuration solutions for these longer pulse sources as described, for example in Ser. No. 10/927,374 (incorporated by reference herein) are different than those described here for sub-picosecond systems. However, the engineering improvements described here will also be applicable for the longer pulse systems, and the bulk amplifier operated as a regenerative amplifier has increased flexibility.
The first important element for a short pulse regenerative amplifier system is the source of short pulses. Femtosecond mode-locked fiber lasers are a good source of such pulses. Typically the fiber oscillator is low power and needs additional amplification for application as a seed source. Other important needs are pulse compression, wavelength flexibility, dispersion control and fiber delivery.
Therefore, it is an object of the present invention to introduce a modular, compact, widely-tunable, high peak and high average power, low noise ultrafast fiber amplification laser system suitable for a seed source for a regenerative amplifier.
It is a further object of the invention to ensure modularity of the system by employing a variety of easily interchangeable optical systems, such as 1) short pulse seed sources, 2) wide bandwidth fiber amplifiers, 3) dispersive pulse stretching elements, 4) dispersive pulse compression elements, 5) nonlinear frequency conversion elements and 6) optical components for fiber delivery. In addition, any of the suggested modules can be comprised of a subset of interchangeable optical systems.
It is a further object of the invention to ensure system compactness by employing efficient fiber amplifiers, directly or indirectly pumped by diode lasers as well as highly integrated dispersive delay lines. The high peak power capability of the fiber amplifiers is greatly expanded by using parabolic or other optimized pulse shapes. In conjunction with self-phase modulation, parabolic pulses allow for the generation of large-bandwidth high-peak power pulses, as well as for well-controlled dispersive pulse stretching. High power parabolic pulses are generated in high-gain single or multi-mode fiber amplifiers operating at wavelengths where the fiber material dispersion is positive.
Parabolic pulses can be delivered or transmitted along substantial fiber lengths even in the presence of self-phase modulation or general Kerr-effect type optical nonlinearities, while incurring only a substantially linear pulse chirp. At the end of such fiber delivery or fiber transmission lines, the pulses can be compressed to approximately their bandwidth limit.
Further, the high energy capability of fiber amplifiers is greatly expanded by using chirped pulse amplification in conjunction with parabolic pulses or other optimized pulse shapes, which allow the toleration of large amounts of self-phase modulation without a degradation of pulse quality. Highly integrated chirped pulse amplification systems are constructed without compromising the high-energy capabilities of optical fibers by using fiber-based pulse stretchers in conjunction with bulk-optic pulse compressors (or low nonlinearity Bragg gratings) or periodically poled nonlinear crystals, which combine pulse compression with frequency-conversion.
The dispersion in the fiber pulse stretcher and bulk optic compressor is matched to quartic order in phase by implementing fiber pulse stretchers with adjustable 2nd, 3rd and 4th order dispersion. Adjustable higher-order dispersion can be obtained by using high numerical aperture single-mode fibers with optimized refractive index profiles by itself or by using standard step-index high numerical aperture fibers in conjunction with linearly chirped fiber gratings. Alternatively, higher-order dispersion can be controlled by using the dispersive properties of the higher-order mode in a high numerical aperture few-moded fiber, by using nonlinearly chirped fiber gratings or by using linearly chirped fiber gratings in conjunction with transmissive fiber gratings. Adjustable 4th order dispersion can be obtained by controlling the chirp in fiber Bragg gratings, transmissive fiber gratings and by using fibers with different ratios of 2nd, 3rd and 4th order dispersion. Equally, higher-order dispersion control can be obtained by using periodically poled nonlinear crystals.
The fiber amplifiers are seeded by short pulse laser sources, preferably in the form of short pulse fiber sources. For the case of Yb fiber amplifiers, Raman-shifted and frequency doubled short pulse Er fiber laser sources can be implemented as widely tunable seed sources. To minimize the noise of frequency conversion from the 1.5 μm to the 1.0 μm regime, self-limiting Raman-shifting of the Er fiber laser pulse source can be used. Alternatively, the noise of the nonlinear frequency conversion process can be minimized by implementing self-limiting frequency-doubling, where the center wavelength of the tuning curve of the doubling crystal is shorter than the center wavelength of the Raman-shifted pulses.
The process of Raman-shifting and frequency-doubling can also be inverted, where an Er fiber laser is first frequency-doubled and subsequently Raman-shifted in an optimized fiber providing soliton-supporting dispersion for wavelengths around 800 nm and higher to produce a seed source for the 1 μm wavelength regime.
As an alternative low-complexity seed source for an Yb amplifier, a modelocked Yb fiber laser can be used. The fiber laser can be designed to produce strongly chirped pulses and an optical filter can be incorporated to select near bandwidth-limited seed pulses for the Yb amplifier.
Presently the mode-locked Yb: doped fiber laser is the preferred oscillator. The preferred source is described Ser. No. 10/627,069 (incorporated herein).
The present invention is similarly directed to a mass-producible passively modelocked fiber laser. By incorporating apodized fiber Bragg gratings, integrated fiber polarizers and concatenated sections of polarization-maintaining and non-polarization-maintaining fibers, a fiber pig-tailed, linearly polarized output can be readily obtained from the laser. By further matching the dispersion value of the fiber Bragg grating to the inverse, or negative, of the dispersion of the intra-cavity fiber, the generation of optimally short pulses with a large optical bandwidth can be induced. In this regard, either positive dispersion in conjunction with negative dispersion fiber gratings or negative dispersion in conjunction with positive dispersion fiber gratings can be implemented. Preferably, the dispersion characteristics of the fiber Bragg grating and the dispersion characteristics of the rest of the intra-cavity elements are matched to within a factor of three. Even more preferably, these characteristics are matched within a factor of two, or within a factor in the range of 1.0 to 2.0. Also preferably, the Bragg grating has a chirp rate greater than 80 nm/cm. More preferably, the Bragg grating has a chirp rate greater than 160 nm/cm. Most preferably, the Bragg grating has a chirp rater greater than 300 nm/cm. To maximize the output power and the pulse repetition rate, the use of wide-bandwidth fiber Bragg gratings with low absolute dispersion is preferable. These fiber Bragg gratings are also used as end-mirrors for the cavity and allow the transmission of pump light to the intra-cavity gain fiber. The fiber Bragg gratings are conveniently produced using phase masks.
Alternatively, fiber couplers can be used inside the fiber cavity. Generally, sections of polarization-maintaining and non-polarization-maintaining fiber can be concatenated inside the fiber cavity. The non-polarization-maintaining section should then be short enough so as not to excessively perturb the polarization state. Intra-cavity sections of non-polarization-maintaining fiber preferably comprise all-fiber polarizers to lead to preferential oscillation of one linear polarization state inside the cavity. Similarly, when directly concatenating polarization-maintaining fiber sections, the length of the individual section should be long enough to prevent coherent interactions of pulses propagating along the two polarization axes of the polarization-maintaining fibers, thereby ensuring a maximum in pulse stability.
Saturable absorber mirrors (SAMs) placed inside the cavity enable passive modelocking. The saturable absorbers (SA) can be made from multiple quantum wells (MQW) or bulk semiconductor films. These saturable absorbers have preferably a bi-temporal life-time with a slow component (>>100 ps) and a fast component (<<20 ps). The realization of the bi-temporal dynamics of the optical nonlinearity is achieved by tailoring the depth profile of the ion-implantation in combination with the implantation dose and energy. The result is that the carriers trap at distinctively different rates in different depth regions of the SAM.
Saturating semiconductor films can for example be grown from aluminum-containing material such as AlGaInAs, the exact composition can be selected depending on the sought band-gap (typically selected to be in the vicinity of the desired operating wavelength of the laser system) and it is also governed by the requirement of lattice-match between the saturating semiconductor film and an underlying Bragg mirror or any other adjacent semiconductor material. Compositional requirements enabling lattice match between semiconductors and/or a certain band gap are well known in the state of the art and are not further explained here.
In aluminum containing semiconductors the surface area can induce a low optical damage threshold triggered by oxidization of the surface. In order to prevent optical damage of aluminum containing surface areas a passivation layer, e.g., InP, InGaAs or GaAs, is incorporated. SA degradation is further minimized by optimizing the optical beam diameter that impinges on the SAM. In one implementation the SAM and an intra-cavity fiber end can be either butt-coupled or brought into close contact to induce modelocking. Here, the incorporation of a precision AR-coating on the intra-cavity fiber end minimizes any bandwidth restrictions from etalon formation between the SAM and the fiber end. Etalons can also be minimized by appropriate wedging of the fiber ends. The beam diameter inside the SAM can be adjusted by implementing fiber ends with thermally expanded cores. Alternatively, focusing lenses can be directly fused to the fiber end. Moreover, graded-index lenses can be used for optimization of the focal size and working distance between the fiber tip and SA surface.
Wavelength tuning of the fiber lasers can be obtained by heating, compression or stretching of fiber Bragg gratings or by the incorporation of bulk optic tuning elements.
The use of bi- or multi-temporal saturable absorbers allows the design of dispersion compensated fiber laser operating in a single-polarization state, producing pulses at the bandwidth limit of the fiber gain medium.
Further improvement of the femtosecond Yb doped fiber oscillator can include an integral mass produced master oscillator, power amplifier design (MOPA) which is describe in Ser. No. 10/814,502 (incorporated by reference herein).
One embodiment of the present invention comprises a master oscillator power amplifier comprising a mode-locked fiber oscillator and a fiber amplifier. The mode-locked fiber oscillator comprises a pair of reflective optical elements that form an optical resonator. At least one of the reflective optical elements is partially transmissive and has a reflection coefficient that is less than about 60%. The mode-locked fiber oscillator outputs a plurality of optical pulses. The fiber amplifier is optically connected to the mode-locked fiber oscillator through a bi-directional optical connection such that light from the mode-locked fiber oscillator can propagate to the fiber amplifier and light from the fiber amplifier can propagate to the mode-locked fiber oscillator.
Another embodiment of the present invention comprises a method of producing laser pulses. In this method, optical energy is propagated back and forth through a gain fiber by reflecting light from a pair of reflective elements on opposite ends of the gain fiber. Less than about 60% of the light in the gain fiber is reflected back into the gain fiber by one of the reflectors. The pair of reflective elements together form a resonant cavity that supports a plurality of resonant optical modes. The resonant optical modes are substantially mode-locking to produce a train of pulses. The train of optical pulses is propagated from the laser cavity through one of the reflectors to a fiber amplifier along a bi-directional optical path from the laser cavity to the fiber amplifier where the laser pulses are amplified.
Another embodiment of the present invention comprises a fiber-based master oscillator power amplifier comprising a mode-locked fiber oscillator, a fiber amplifier comprising a gain fiber, and bi-directional optical path between the mode-locked fiber oscillator and the fiber amplifier. The mode-locked fiber oscillator comprises a resonant cavity and a gain medium. The mode-locked fiber oscillator produces a plurality of optical pulses. The bi-directional optical path between the mode-locked fiber oscillator and the fiber amplifier permits light from the mode-locked fiber oscillator to propagate to the fiber amplifier and light from the fiber amplifier to propagate to the mode-locked fiber oscillator. The mode-locked fiber oscillator comprises a first segment of fiber and the fiber amplifier comprises a second segment of optical fiber. The first and second segments form a substantially continuous length of optical fiber. In some embodiments, the first and second segments are spliced together. The first and second segments may be fusion spliced. The first and second segments may also be butt coupled together with or without a small gap, such as a small air gap, between the first and second segments.
Another embodiment of the present invention comprises a method of producing laser pulses comprising substantially mode-locking longitudinal modes of a laser cavity to produce laser pulses and propagating the laser pulses from the laser cavity to a fiber amplifier. The laser pulses are amplified in the fiber amplifier. Amplified spontaneous emission emitted from the fiber amplifier is received at the laser cavity. A first portion of the spontaneous emission enters the laser cavity. A second portion of the amplified spontaneous emission from the laser cavity is retro-reflected back to the fiber amplifier to cause the second portion to be directed away from the cavity toward the fiber amplifier.
Another embodiment of the present invention comprises a fiber master oscillator power amplifier comprising a mode-locked fiber oscillator and a fiber amplifier. The mode-locked fiber oscillator comprises a first portion of optical fiber and a pair of reflectors spaced apart to form a fiber optic resonator in the first fiber portion. At least one of the fiber reflectors comprises a partially transmissive fiber reflector. The mode-locked fiber oscillator outputs a plurality of optical pulses. The fiber amplifier comprises a second portion of optical fiber optically connected to the partially transmissive fiber reflector to receive the optical pulses from the mode-locked oscillator. The second portion of optical fiber has gain to amplify the optical pulses. The first portion of optical fiber, the partially transmissive fiber reflector, and the second portion of optical fiber comprise a continuous path formed by optical fiber uninterrupted by non-fiber optical components.
Another embodiment of the present invention comprises a master oscillator power amplifier comprising a mode-locked fiber oscillator and a fiber amplifier. The mode-locked fiber oscillator comprises a pair of reflective optical elements that form an optical resonator. At least one of the reflective optical elements comprises a partially transmissive Bragg fiber grating having a reflection coefficient that is less than about 60%. The mode-locked fiber oscillator outputs a plurality of optical pulses. A fiber amplifier is optically connected to the oscillator through an optical connection to the partially transmissive Bragg fiber grating.
Another embodiment of the present invention comprises a master oscillator power amplifier comprising a mode-locked fiber oscillator, a fiber amplifier, and a pump source. The mode-locked fiber oscillator comprises a pair of reflective optical elements that form an optical resonator. At least one of the reflective optical elements is partially transmissive and has a reflection coefficient that is less than about 60%. The mode-locked fiber oscillator outputs a plurality of optical pulses. A fiber amplifier is optically connected to the oscillator through an optical connection to the at least one partially transmissive reflective optical elements. The pump source is optically connected to the mode-locked fiber oscillator and the fiber amplifier to pump the mode-locked fiber oscillator and the fiber amplifier.
However, for most embodiments for a source for a regenerative amplifier the pulses need to be conditioned before amplification. Ser. No. 10/814,319 (incorporated by reference herein) addresses the utilization of modules so that the correct performance can be obtained from the femtosecond source for the seeder or a portion of the seeder for the regenerative amplifier system. Parameter controls available through these modules can be utilized for the optimization of the output from the regenerative amplifier.
One embodiment of the invention thus comprises a pulsed fiber laser outputting pulses having a duration and corresponding pulse width. The pulsed laser comprises a modelocked fiber oscillator, an amplifier, a variable attenuator, and a compressor. The modelocked fiber oscillator outputs optical pulses. The amplifier is optically connected to the modelocked fiber oscillator to receive the optical pulses. The amplifier comprises a gain medium that imparts gain to the optical pulse. The variable attenuator is disposed between the modelocked fiber oscillator and the amplifier. The variable attenuator has an adjustable transmission such that the optical energy that is coupled from the mode-locked fiber oscillator to the amplifier can be reduced. The compressor compresses the pulse thereby reducing the width of the pulse. Preferably a minimum pulse width is obtained.
Another embodiment of the invention comprises a method of producing compressed high power short laser pulses having an optical power of at least about 200 mW and a pulse duration of about 200 femtoseconds or less. In this method, longitudinal modes of a laser cavity are substantially mode-locked to repetitively produce a laser pulse. The laser pulse is amplified. The laser pulse is also chirped thereby changing the optical frequency of the optical pulse over time. The laser pulse is also compressed by propagating different optical frequency components of the laser pulse differently to produce compressed laser pulses having a shortened temporal duration. In addition, the laser pulse is selectively attenuated prior to the amplifying of the laser pulse to further shorten the duration of the compressed laser pulses.
Another embodiment of the invention comprises a method of manufacturing a high power short pulse fiber laser. This method comprises mode-locking a fiber-based oscillator that outputs optical pulses. This method further comprises optically coupling an amplifier to the fiber-based oscillator through a variable attenuator so as to feed the optical pulses from the fiber-based oscillator through the variable attenuator and to the amplifier. The variable attenuator is adjusted based on a measurement of the optical pulses to reduce the intensity of the optical pulses delivered to the amplifier and to shorten the pulse.
Another embodiment of the invention comprises a pulsed fiber laser outputting pulses having a pulse width. The pulsed fiber laser comprises a modelocked fiber oscillator, an amplifier, and a spectral filter. The modelocked fiber oscillator produces an optical output comprising a plurality of optical pulses having a pulse width and a spectral power distribution having a bandwidth. The amplifier is optically connected to the modelocked fiber amplifier for amplifying the optical pulses. The spectral filter is disposed to receive the optical output of the modelocked fiber oscillator prior to reaching the amplifier. The spectral filter has a spectral transmission with a band edge that overlaps the spectral power distribution of the optical output of the modelocked fiber oscillator to attenuate a portion of the spectral power distribution and thereby reduce the spectral bandwidth. The pulse width of the optical pulses coupled from the mode locked fiber oscillator to the fiber amplifier is thereby reduced.
Another embodiment of the invention comprises a method of producing compressed optical pulses. In this method, longitudinal modes of a fiber resonant cavity are substantially mode-locked so as to produce a train of optical pulses having a corresponding spectral power distribution with a spectral bandwidth. The optical pulses are amplified and compressed to produce compressed optical pulses. The spectral bandwidth of the spectral power distribution is reduced such that the compressed optical pulses have a shorter duration.
Another embodiment of the invention comprises a pulsed fiber laser comprising a modelocked fiber oscillator, an amplifier, one or more optical pump sources, a pulse compressor, and a pre-compressor. The modelocked fiber oscillator comprises a gain fiber and a pair of reflective optical elements disposed with respect to the gain fiber to form a resonant cavity. The modelocked fiber oscillator produces a train of optical pulses having an average pulse width. The amplifier is optically connected to the modelocked fiber amplifier such that the optical pulses can propagate through the amplifier. The fiber amplifier amplifies the optical pulses. The one or more optical pump sources are optically connected to the modelocked fiber oscillator and the fiber amplifier to pump the fiber oscillator and fiber amplifier. The pulse compressor is optically coupled to receive the amplified optical pulses output from fiber amplifier. The pulse compressor shortens the pulse width of the optical pulses output by the fiber amplifier. The pre-compressor is disposed in an optical path between the modelocked fiber oscillator and the fiber amplifier. The pre-compressor shortens the duration of the optical pulses introduced into the fiber amplifier such that the pulse duration of the optical pulses output by the compressor can be further shortened.
Another embodiment of the invention comprises a method of generating short high power optical pulses. The method comprises substantially mode-locking optical modes of a laser cavity to produce an optical signal comprising a plurality of laser pulses having an average pulse width. The optical signal comprises a distribution of frequency components. The method further comprises compressing the optical pulses and amplifying the compressed optical pulses to produce amplified compressed optical pulses. The amplified compressed optical pulses are further compressed subsequent to the amplifying using a dispersive optical element to differentiate between spectral components and introducing different phase shifts to the different spectral components.
Another embodiment of the invention comprises a pulsed fiber laser comprising a modelocked fiber oscillator, a fiber amplifier, an optical pump source, and a pulse compressor. The modelocked fiber oscillator outputs optical pulses. The fiber amplifier is optically connected to the modelocked fiber oscillator and amplifies the optical pulses. The optical pump source is optically connected to the fiber amplifier. The pulse compressor is optically coupled to receive the amplified optical pulses output from the fiber amplifier. The pulsed fiber laser further comprises at least one of (i) a first optical tap in the optical path between the modelocked fiber oscillator and the fiber amplifier and a first feedback loop from the first tap to control the modelocked fiber oscillator based on measurement of output from the first optical tap, and (ii) a second optical tap in the optical path between the fiber amplifier and the compressor and a second feedback loop from the second tap to control the fiber amplifier based on measurement of output from the first optical tap.
Another embodiment of the invention comprises a pulsed light source comprising a light source module, an isolator module, an amplifier module, and a compressor module. The light source module comprises an optical fiber and outputs optical pulses. The isolator module comprises an optical isolator in a housing having input and output fibers. The input fiber is optically coupled to the optical fiber of the light source module. The optical isolator is disposed in an optical path connecting the input and output fibers such that the optical pulses introduced into the input fiber are received by the isolator and permitted to continue along the optical path to the output coupler. The amplifier module comprises an amplifying medium and has an optical input optically connected to the output fiber of the isolator module to amplify the optical pulses. The compressor module is optically coupled to the amplifier module to compress the optical pulses.
Up to this point a mode-locked fiber laser or a bulk solid state mode-locked laser as the seed source for the fiber amplifier and regenerative amplifier has been disclosed. Other sources can also be utilized such a laser-diodes or microchip lasers. In Ser. No. 10/437,057 (incorporated by reference herein), it is disclosed how to modify these sources to give higher energy and shorter pulses through amplification and pulse compression in fiber amplifiers. An advantage of these sources that is mentioned in Ser. No. 10/437,057 is the repetition rate can be variable. It is a true advantage to match the repetition rate of the source to that of the regenerative amplifier.
Thus, one object of this invention is to convert relatively long pulses from rep-rate variable ultrafast optical sources to shorter, high-energy pulses suitable for seed sources in high-energy ultrafast lasers including a regenerative amplifier. Another object of this invention is to take advantage of the need for higher pulse energies at lower repetition rates so that such sources can be cost effective.
A gain switched laser diode as is used in telecom systems can be used as the initial source of pulses. In this case, the diode is operated at a much lower repetition rate. The pulses are still amplified in fiber amplifiers. Fiber amplifiers can be used as constant output power devices. The upper-state lifetime in typical doped amplifier fibers such as Ytterbium and Erbium is in the millisecond range so that these amplifiers can amplify pulse trains with the same efficiency at repetition rates from 10's of kHz to 100's of GHz and beyond. If the amplifier is amplifying pulses at 10 kHz rather than at 10 GHz at constant power, then the pulse energy will be six orders of magnitude higher. Again, with such high peak powers, pulse compression methods need to be different and unique. One first embodiment uses conventional compression by spectral broadening the pulses in an optical fiber with positive group velocity dispersion (GVD) and then compressing the pulse with diffraction gratings. The object of the pulse compression is to convert the 3-25 picosecond pulses from the gain switched laser diode to pulses that are subpicosecond.
Another source starts with pulses from a low cost Q-switched microchip laser. These lasers give pulses as short as 50 picoseconds but typically 250 picoseconds to 1.0 nanosecond. The pulse peak powers are typically 1-10 kW with pulse energies 6 orders of magnitude higher than from telecom laser diodes. Microchip lasers could be a very cost effective source for pulses less than 10 picoseconds with suitable pulse compression methods. Single mode fiber compression has thus far been limited to pulses shorter than 150 ps and peak powers less than 1 kW. Before compression the pulse can be further amplified in a regenerative amplifier.
Once a suitable femtosecond source has been identified further improvements have been made in the incorporation of these lasers in chirped pulse amplification systems where the amplifier has been a fiber amplifier. In Ser. No. 10/813,163, many improvements to the fiber chirped pulse amplification (FCPA) configuration have been made for a configuration that is more robust and suitable to an industrial environment. Here it has been realized that these improvements can be also utilized for fiber lasers seeding solid state amplifiers and particular solid state regenerative amplifiers. Specifically, the improvements for the FCPA configuration that are disclosed in Ser. No. 10/813,163 can be utilized in a regenerative amplifier seeded with a fiber laser configuration. The simplest embodiments are for the replacement of the power amplifier in
The following topics that are covered in Ser. No. 10/813,163 are relevant to this configuration.
The invention thus relates to the technologies necessary to overcome the above problems and limitations of the prior art, to build a hybrid fiber and solid-state based chirped pulse amplification laser system suitable for industrial applications, with the fiber in a modular and compact laser design with all modules replaceable. The modules are designed and manufactured to telecom standards and quality.
Environmentally stable laser design is crucial for industrial application. An industrial laser system can be, for example, characterized by an output power variation below 0.5 dB over an environmental temperature range from 0 to 50 degrees Celsius, and by compliance with the vibration, thermal shock, high temperature storage and thermal cycling test criteria in Telcordia GR468-CORE and GR-1221-CORE. This target can be achieved by functional segmentation of the components and packaging the modular device with Telcordia-qualified packaging technology. Before the modules are assembled into a system, they are tested and assembled separately.
Included in the modules are tap units that allow taking out signals along the propagation path in an integrated design. This is necessary for the optimization of each module as it is assembled, and important in the spectral matching along the chain of modules.
Polarization units are provided to prevent the buildup of side-pulses from orthogonal polarization light.
The acousto-optical down counter module can be designed to operate as a bandwidth filter. For further modulation of the signal an additional pulse extractor can be included near the end of the output. This unit suffers from dispersion due to the large bandwidth of the pulse. The compressor can be used to correct for this dispersion as disclosed hereafter.
The invention also relates to a means to extract one or more chirped pulses from a series of such pulses using an acousto-optic deflector, and to compensate for the detrimental effects on the spatial characteristics of the extracted chirped pulse caused by dispersion in that deflector. An important aspect of this system is to manage the spectrum of the pulse in the system while maintaining the ability to correct for dispersion and compress the pulse back to the femtosecond regime. Two principal embodiments of this type will be described. The first is the case where the spectral content of the seed pulse is small. In this case a nonlinear amplifier may be employed for the generation of additional spectrum while spectral filtering is employed to obtain a compressible pulse. The second case is where the spectrum from the source is larger than necessary. Nonlinear affects can be limited in the amplifier chain in this case, while spectral filtering is again employed to obtain a compressible pulse. An additional attribute that is necessary for many applications is the reduction of the ASE at the output. Specific amplifier designs are used to cut the ASE at the output wavelength. The compressor can be used as an optical spectral filter to this end.
Once gain performance is attained, a method for active stabilization of the optical performance of the gain fiber in a laser or amplifier is disclosed to maintain this performance. The present invention stabilizes the temperature dependent absorption of a gain fiber over a wide environmental temperature variation by an active feedback loop. A piece of fiber, optically identical with the gain fiber itself, is used as a spectral filter for monitoring the emission spectrum of the pump diode. The absorption spectrum of the filter fiber follows that of the gain fiber if both fibers are packaged so that the fibers are in proximity to each other. The transmission of the pump light through the filter fiber clones exactly the absorption characteristics of the gain fiber at a given package temperature. The temperature of the pump diode is controlled by a feedback loop such that the transmission through the filter fiber is maintained at the minimum. Importantly, the filter fiber functions as an active temperature sensor of the gain fiber. Precise spectral control of the gain at any fiber or package temperature can thus be realized.
As mentioned above, an important field of use for this system is in micromachining. An additional feature needed for this application field is the capability to start and stop the pulse stream while moving the targeted material in place. One method to do this is to control the down counter. However, this leads to problems with gain stabilization in the amplifier and excessive ASE on target. These problems have been addressed in Ser. No. 10/813,173 “Method and Apparatus for Controlling and Protecting Pulsed High Power Fiber Amplifier Systems” (incorporated by reference herein). However, another means to stop the pulse stream is to utilize an optical switch at the output.
The invention extracts one or more chirped pulses from a series of such pulses using an acousto-optic deflector, and compensates for the detrimental effects on the spatial characteristics of the extracted chirped pulse caused by dispersion in that deflector. The instant invention has the additional advantage that the means to compensate for dispersion in the acousto-optic deflector can be used to compress the duration of the chirped pulse. This is accomplished by placing the AOM in proximity to a grating compressor.
Further improvements for correction of higher order dispersion terms in fiber chirped pulse amplification systems are disclosed in Ser. No. 10/992,762 (incorporated by reference herein). These can be applied to chirped pulse amplification systems with regenerative amplifiers.
Here, an ultra-compact high energy chirped pulse amplification systems based on linearly or nonlinearly chirped fiber grating pulse stretchers and photonic crystal fiber pulse compressors. Alternatively, photonic crystal fiber pulse stretchers and photonic crystal fiber compressors can also be implemented. For industrial applications the use of all-fiber chirped pulse amplification systems is preferred, relying on fiber-based pulse compressors and stretchers as well as fiber-based amplifiers.
Fiber-based high energy chirped pulse amplification systems of high utility can also be constructed from conventional optical components such as pulse stretchers based on long lengths of conventional fiber as well as bulk grating compressors. The performance of such ‘conventional’ chirped pulse amplification systems can be greatly enhanced by exploiting nonlinear cubicon pulse formation, i.e. by minimization of higher-order dispersion via control of self-phase modulation inside the amplifiers.
Finally, a particularly compact seed source for an Yb fiber-based chirped pulse amplification system can be constructed from an anti-Stokes frequency shifted modelocked Er fiber laser amplifier system, where a wavelength tunable output is obtained by filtering of the anti-Stokes frequency shifted output. The noise of such an anti-Stokes frequency shifted source is minimized by the amplification of positively chirped pulses in a negative dispersion fiber amplifier.
The preceding improvements have been focused on systems operating close to 1 μm. These systems appear to be the most suitable for industrial applications. However, Ti:sapphire regenerative amplifiers are presently the dominant design. Frequency doubled erbium fiber lasers are utilized for the more industrial Ti:sapphire systems. FCPA front ends are suitable for higher repetition rates utilizing an electro-optic pulse selector as is disclosed in Ser. No. 10/960,923. FCPA systems operating in the 1.5 telecomm wavelength which are then frequency doubled would be suitable for a Ti:sapphire amplifier or regenerative amplifier system.
The invention in Ser. No. 10/606,829 (incorporated by reference herein) provides an erbium fiber (or erbium-ytterbium) based chirped pulse amplification system operating at a wavelength of approximately 1550 nanometers. The use of fiber amplifiers operating in the telecommunications window enables telecommunications components and telecommunications compatible assembly procedures to be used, with superior mechanical stability
It is found that electronic controls are needed for reliable operation for these complex systems. In Ser. No. 10/813,173 (incorporated by reference herein), the implementation of electronic controls are described which prevent catastrophic damage in a short pulse amplifier system as well as maintaining constant output power over the life of the system. These systems are very applicable in a regenerative amplifier system seeded by a fiber laser. The damage issues will also be a concern in a regenerative amplifier system. However, more importantly these front end systems normally will encompass nonlinear optical processes in the fiber amplifiers. These nonlinear optical processes are very dependent on laser intensity. Thus, to maintain the desired results over the life of the system, careful control of the optical powers is needed particularly in the nonlinear optical components in the system.
It is thus an object of the present invention to provide a high power fiber amplifier system with means for controlling the pump diode current and the gain of the fiber amplifier such that the output pulse energy is constant as the pulse width and repetition rate are adjusted during operation. This includes keeping the pulse energy constant during turn-on of the pulse train.
It is a further object of the invention to provide means for controlling the temperature of the fiber amplifier pump diode such that the pump diode wavelength is maintained at a fixed value with changes in diode current.
It is also an object of the invention to provide means for protecting the high power amplifier from damage due to gain buildup in excess of the damage threshold of the amplifier by monitoring the repetition rate of the injected oscillator pulses or external signal, and shutting off or reducing the pump diode current if the repetition rate falls below this threshold.
It is also an object of the invention to provide for monitoring of the amplitude of the seed pulses and to protect the high power fiber amplifier from damage by shutting off the pump diode if the amplitude of the injected pulses falls outside a safe threshold.
It is also an object of the invention to provide a high power amplifier system with means for controlling the amplitude of the seed pulse such that the output energy of the power amplifier is constant.
The above and other objects of the invention are met by providing a device and method for controlling the diode current of the pump diode in a high power fiber amplifier, the device comprising a means for setting the pump diode current or power, monitoring such current or power, and maintaining the diode current or power at a constant value. Typically the current of the diode is controlled to correct for long term decrease on its output due to aging. In contrast, in accordance with an embodiment of the present invention, the pump diode current is controlled to dynamically control the gain of the power amplifier to maintain uniform pulse energy as the repetition rate and the pulse temporal width is changed. This includes turning the pump diode on sufficiently in advance and ramping up the current to produce equal power for the first pulses when the unit is turned on.
The device also provides a means for calculating and/or storing the desired pump diode current setting as a function of system pulse width and repetition rate, such that the energy of the output pulse is maintained at a desired value as the pulse width and repetition rate are varied.
A device in accordance with an embodiment of the invention also provides a means for calculating and storing the appropriate pump diode temperature setting as a function of the pump diode current setting, such that the emission wavelength of the pump diode is maintained at a wavelength that provides maximum absorption of the pump diode energy by the fiber amplifier medium as the pump diode current is varied.
Means are also provided to monitor the repetition rate of the injected pulses into the amplifier system, to compare it to the predetermined repetition rate, and if lower than this repetition rate, to disable or reduce the current to the amplifier pump diode to prevent it from being damaged.
The exemplary device discussed above also provides a means for comparing the amplitude of the pulse being injected into the fiber amplifier with a predetermined minimum amplitude value and if lower than this predetermined minimum, a means to disable or reduce the current to the amplifier pump diode to prevent it from being damaged. A device in accordance with an embodiment of the invention also provides a means of selecting and attenuating the seed pulses such that the amplified output pulses are of uniform energy.
It is an even further object of the invention to monitor the repetition rate of the oscillator and to provide a means for calculating the required down counter divide ratio needed to obtain a lower repetition rate.
It is also an object of the invention to synchronize the oscillator with an external reference signal. It is also an object of the invention to vary this external reference in frequency, and have the oscillator repetition rate vary accordingly.
It is an even further object of the invention to vary the external reference in frequency, and have the oscillator repetition rate vary accordingly, and also have the down counted repetition rate vary accordingly. However, this variation will be of a limited range compared to an all fiber system due to the operation repetition rate of a regenerative amplifier.
Finally, these regenerative amplifier systems will be utilized in many cases for micromachining. Improvements for FPCA systems have been developed that are unique for a fiber seed source. Ser. No. 10/813,389 (incorporated by reference herein), describes the benefit for changing the pulse shapes that allow the change of the material processing properties of that laser. These methods include allowing the addition of heat by the addition of longer pulses. The physical means for changing these pulse shapes and building a all fiber chirped pulse amplification system suitable for material processing is described Ser. No. 10/813,269 (incorporated by reference herein). As is mentioned in Ser. No. 10/813,269 some of these changes in the seed source for a fiber chirped pulse amplification systems will also be suitable for regenerative amplifier systems. Herein further methods of obtaining various pulse changes are described.
The invention thus provides methods of materials processing using bursts of laser light comprised of ultrashort pulses in the femtosecond, picosecond and nanosecond ranges, wherein parameters of the pulses comprising the burst, such as pulse width, pulse separation duration, pulse energy, wavelength and polarization, are manipulated to induce desirable properties in the processed material.
While a precise and controlled removal of material is achieved using ultrashort pulses, there are situations when having a small amount of thermal effect retained by the material from the previous pulse prior to being irradiated by a subsequent ultrashort pulse is beneficial. In addition, it is well known that the properties of most materials have some dependence on temperature. For example, the absorption of light by silicon is very dependent on temperature. Hence, heating such a target material can help initiate the ablation process at lower threshold fluence and may produce a smoother surface. In general, the thermal and physical effect or any change in structure caused by the prior pulse influences the laser matter interaction with the next pulse.
The ablation threshold energy density, as a function of pulse width, can vary significantly from the square root of t as pulse widths enter the femtosecond range. These ultrashort pulses can be used to micro-machine cleanly without causing significant heat. These ultrashort pulses also have deterministic thresholds compared to the statistical thresholds of longer pulses.
The present invention may be used in micro-machining with bursts of pulses having pulse shapes that cannot be quantified by a single pulse width in order to describe their micro-machining properties. For example, a burst comprises a 100 femtosecond pulse and a one nanosecond pulse, where the one nanosecond pulse contains ninety percent of the energy and the 100 femtosecond pulse contains ten percent of the energy. The threshold for ablation of gold is a little over 0.3 J/cm2 for the 100 femtosecond pulse and 3.0 J/cm2 for the one nanosecond pulse. Thus, if the burst is focused to output 0.3 J/cm2, then ablation will occur during the 100 femtosecond pulse, and not during the one nanosecond pulse. If the one nanosecond pulse impinges upon the surface first, it will have no affect while the 100 femtosecond pulse will ablate. Thus, the one nanosecond predominant pulse will not leave a heat affected zone. However, if the 100 femtosecond pulse is right before the one nanosecond pulse, then the 100 femtosecond pulse will change the absorption properties of the material so the one nanosecond pulse will also interact with the material. In this case, the ablation process would be predominantly heat related. If the one nanosecond pulse is increased to 100 nanoseconds, then the pulse energy content in the long pulse can be increased by ten-fold but the threshold is still determined by the ultrashort pulse and remains fixed even with one percent of the total energy in the ultrafast pulse.
Thus, in one embodiment of the present invention, the long pulse is before the ultrafast pulse if the pulse repetition rate is substantially greater than or equal 100 kilohertz. In another embodiment of the present invention, a portion of the long pulse follows after the ultrafast pulse, and adding a pedestal on the short pulse can create the long pulse. Micro-machining can be accomplished with an ultrashort pulse, where substantial energy is in a long pulse pedestal (>ten picoseconds) and where the long pulse pedestal adds a thermal machining mechanism.
The present invention can perform laser machining on material using a burst of ultrashort laser pulses and tailors the pulse width, pulse separation duration, wavelength and polarization to maximize the positive effect of thermal and physical changes achieved by the previous pulse on the laser matter interaction in a burst-machining mode. Better processing results can be achieved by manipulating the pulse width, the pulse separation duration and the pulse energies of pulses within a burst. The wavelength and polarization of a laser beam also strongly affect the absorption of the laser beam, and have to be varied pulse-to-pulse in a burst in order to produce maximum laser-matter interaction.
Besides the methods of manipulating laser beam parameters described above to achieve desired results, the present invention also includes methods to achieve the thermal and physical enhancement of a material during laser processing. In an embodiment of the present invention, the background light (commonly referred to as Amplified Spontaneous Emission (ASE)) is controlled to provide a constant source of energy for achieving thermal and physical changes to enhance the machining by individual ultrashort pulses. ASE is often emitted simultaneously and co-linearly with the ultrashort pulse from an amplified fiber laser. There are a number of ways to change the ASE ratio in the laser. Examples are changing the ultrashort pulse input energy into the amplifier, changing its center wavelength or changing the diode pump power to the amplifier. Another means more variable is within the compressor of the laser. As disclosed in application Ser. No. 10/813,163 the spectral output of the ASE can be designed to be at a different wavelength then that of the ultrashort pulse. Thus, in the compressor, where the spectral components are physically separated, a component can be placed to block or partially block the ASE, as disclosed in application Ser. No. 10/813,163. The ratio between the ASE and the ultrashort pulse energy can be controlled to vary the amount of preheating applied to the target material. In another embodiment of the invention, a pedestal of an ultrashort pulse is controlled. The pedestal is similar to a superimposed long-pulse with lower amplitude.
The invention is based on the interaction with a material of laser pulses of different pulse widths, pulse separation duration, energy, wavelength and polarization in a burst mode. The positive aspects of pulses having different pulse widths, pulse separation duration, energies, wavelengths and polarization are utilized, and a negative aspect of one pulse complements a positive aspect of another pulse. The coupling of laser energy during interaction of successive laser pulses with a material induces various thermal, physical and chemical couplings. The induced coupling involves microscopic change of electronic structure, phase transition, structural disintegration and/or other physical changes. For example, pulses with different pulse widths in a burst induce coupling that is different from a burst having pulses with the same pulse width.
An aspect of the invention provides a method of materials processing using laser light. The method comprises applying bursts of laser light to a target area of a material at a predetermined repetition rate. Preferably, the burst repetition rate is large enough for multipulse pulses generated within the round trip time of the regenerative amplifier, although lower repetition rates can be used. The burst of laser light comprises a first pulse and a second pulse of laser light displaced in time, although more pulses could be used in the burst as necessary. The first pulse has a first pulse width and the second pulse has a second pulse width, and predetermined parameters of the first pulse are selected to induce a change in a selected property of the processed material. The second pulse has a second pulse width, and predetermined parameters of the second pulse are selected based upon the property change induced by the first pulse. The first pulse width is generally in the nanosecond range, and the second pulse width is generally in the picosecond to femtosecond range. However, as stated previously it can be reversed. Predetermined parameters include pulse energy, pulse wavelength, pulse separation duration and pulse polarization vector. These parameters of the first and second pulses are controlled as well to machine the target area of the processed material.
A still further aspect of the present invention provides a method of materials processing that is similar to the previous aspect, except that the first and second pulses of the burst of laser light are overlapped in time, instead of being displaced in time. More pulses could be used in the burst as necessary. The first pulse has a first pulse width and the second pulse has a second pulse width, and the first pulse width can be greater than the second pulse width. The first pulse has a first pulse width and predetermined parameters of the first pulse are selected to induce a change in a selected property of the processed material. The second pulse has a second pulse width, and predetermined parameters of the second pulse are selected on based upon the property change induced by the first pulse. The first pulse width is generally in the nanosecond range, and the second pulse width is generally in the picosecond to femtosecond range. Predetermined parameters include pulse energy, pulse wavelength, pulse separation duration and pulse polarization vector which are controlled as well to machine the target area of the processed material. In addition, the second pulse may include a pedestal to facilitate thermally heating the processed material.
In yet another aspect of the present invention, an apparatus for generating optical pulses, wherein each pulse may have individualized characteristics, is provided. The apparatus comprises a laser means for generating the bursts of pulses, a control means that controls the laser means and a beam manipulation means for monitoring the pulse width, wavelength, repetition rate, polarization and/or temporal delay characteristics of the pulses comprising the pulse bursts. The apparatus generates feedback data based on the measured pulse width, wavelength, repetition rate, polarization and/or temporal delay characteristics for the control means. In one embodiment of the present invention, the laser means may comprise a fiber amplifier that uses stretcher gratings and compressor gratings. The beam manipulation means can comprise a variety of devices, e.g., an optical gating device that measures the pulse duration of the laser pulses, a power meter that measures the power of the laser pulses output from the laser means or a photodiode that measures a repetition rate of the laser pulses. Another beam manipulation means optically converts the fundamental frequency of a percentage of the generated laser pulses to one or more other optical frequencies, and includes at least one optical member that converts a portion of the fundamental of the laser pulses into at least one higher order harmonic signal. The optical member device may comprise a non-linear crystal device with a controller that controls the crystal's orientation. Preferably, the means for converting an optical frequency includes a spectrometer that measures predetermined parameters of pulses output from the non-linear crystal device and generates feedback for the control means.
The accompanying drawings, which are incorporated in and constitute a part of this specification illustrate embodiments of the invention and, together with the description, serve to explain the aspects, advantages and principles of the invention. In the drawings,
a is a diagram of a saturable absorber mirror according to an embodiment of the invention.
b is a diagram of a saturable absorber mirror according to an alternative embodiment of the invention.
a is a diagram of a scheme for coupling a saturable absorber mirror to a fiber end according to an embodiment of the invention.
b is a diagram of a scheme for coupling a saturable absorber mirror to a fiber end according to an alternative embodiment of the invention.
a is a diagram of a passively modelocked fiber laser based on concatenated sections of polarization maintaining and non-polarization maintaining fiber sections according to an embodiment of this invention.
b is a diagram of a passively modelocked fiber laser based on concatenated sections of long polarization maintaining fiber sections according to an embodiment of this invention.
c is a diagram of a passively modelocked fiber laser based on short concatenated sections of polarization maintaining fiber and additional sections of all-fiber polarizer according to an embodiment of this invention.
a illustrates a module usable for polarization correction or as variable attenuation in a fiber based laser system.
b illustrates a particularly preferred embodiment for a fiber solid-state regenerative amplifier system.
a-43c show outputs of the pulse source of
A generalized illustration of the system of the invention is shown in
A generalized illustration of one embodiment of the short pulse source 11 is shown in
The embodiment shown in
In the following, discussion is restricted to the SM-PSM-AM1-PCM system. The SM 1 preferably comprises a femtosecond pulse source (seed source 5). The PSM preferably comprises a length of fiber 6, where coupling between the SM and the PSM is preferably obtained by fusion splicing. The output of the PSM is preferably injected into the fundamental mode of the Yb amplifier 7 inside the AM1 module 3. Coupling can be performed by fusion splicing, a fiber coupler or a bulk-optic imaging system between PSM 2 and the fiber amplifier 7. All fibers are preferably selected to be polarization maintaining. The PCM 4 is preferably a dispersive delay line constructed from one or two bulk optic diffraction gratings for reasons of compactness. Alternatively, a number of bulk optic prisms and Bragg gratings can be used inside the PCM 4. Coupling to the PCM 4 can be performed by a bulk optic lens system as represented by the single lens 8 in
As an example of a femtosecond laser seed source, a Raman-shifted, frequency-doubled Er fiber laser is shown within an SM 1b in
For optimum Raman-shifting from 1.5 μm to the 2.1 μm wavelength region, a reduction in the core diameter (tapering) along the length of the polarization maintaining Raman-shifting fiber 10 is introduced. A reduction of the core diameter is required to keep the 2nd order dispersion in the Raman-shifter close to zero (but negative) in the whole wavelength range from 1.5 to 2.1 μm. By keeping the absolute value of the 2nd order dispersion small, the pulse width inside the Raman shifter is minimized, which leads to a maximization of the Raman frequency shift (J. P. Gordon, “Theory of the Soliton Self-frequency Shift,” Opt. Lett., 11, 662 (1986)). Without tapering, the Raman frequency-shift is typically limited to around 2.00 μm, which even after frequency-doubling is not compatible with the gain bandwidth of Yb fiber amplifiers.
In this particular example, a two-stage Raman shifter 10 consisting of 30 and 3 m lengths of silica ‘Raman’ fiber (single-mode at 1.56 μm) with core diameters of 6 and 4 μm respectively, was implemented. Due to the onset of the infrared absorption edge of silica at 2.0 μm, it is beneficial to increase the rate of tapering towards the end of the Raman shifter 10. In the present example, conversion efficiencies up to 25% from 1.57 μm to 2.10 μm were obtained. Even better conversion efficiencies can be obtained by using a larger number of fibers with smoothly varying core diameter, or by implementing a single tapered fiber with smoothly varying core diameter.
Frequency-conversion of the Raman-shifted pulses to the 1.05 μm region can be performed by a length of periodically poled LiNbO3 (PPLN) crystal 11 with an appropriately selected poling period. (Although throughout this specification, the preferable material for frequency conversion is indicated as PPLN, it should be understood that other periodically-poled ferroelectric optical materials such as PP lithium tantalate, PP MgO:LiNbO3, PP KTP, or other periodically poled crystals of the KTP isomorph family can also be advantageously used.) Coupling with the PPLN crystal 11 occurs through the use of a lens system, represented in
In addition, when the frequency-converted pulse width is substantially longer than the pulse width of the Raman-shifted pulses, the wide bandwidth of the Raman-pulses can be exploited to allow for wavelength-tuning of the frequency-converted pulses, i.e., efficient frequency conversion can be obtained for pulses ranging in frequency from 2(ω1−δω) to 2(ω1+δω), where 2δω is the spectral width at half maximum of the spectrum of the Raman-shifted pulses. Continuous wavelength tuning here is simply performed by tuning the temperature of the frequency-conversion crystal 11.
The amplified noise of the Raman-shifter, PPLN-crystal combination is minimized as follows. Self-limiting Raman-shifting of the Er fiber laser pulse source can be used by extending the Raman shift out to larger than 2 μm in silica-based optical fiber. For wavelengths longer than 2 μm, the infrared absorption edge of silica starts to significantly attenuate the pulses, leading to a limitation of the Raman shift and a reduction in amplitude fluctuations, i.e., any increase in pulse energy at 1.5 μm tends to translate to a larger Raman-shift and thus to a greater absorption in the 2 μm wavelength region, which thus stabilizes the amplitude of the Raman-shifted pulses in this region.
Alternatively, the noise of the nonlinear frequency conversion process can be minimized by implementing self-limiting frequency-doubling, where the center wavelength of the tuning curve of the doubling crystal is shorter than the center wavelength of the Raman-shifted pulses. Again, any increase in pulse energy in the 1.5 μm region translates into a larger Raman-shift, producing a reduced frequency conversion efficiency, and thus the amplitude of the frequency-doubled pulses is stabilized. Therefore a constant frequency-converted power can be obtained for a large variation in input power.
This is illustrated in
Several options exist for the PSM 2. When a length of fiber 6 (stretching fiber) is used as a PSM as shown in
For pulse stretching by more than a factor of 10, the control of third-order and higher-order dispersion becomes important for optimal pulse compression in the PCM 4. To overcome this problem, the stretcher fiber 6 in the PSM 2 can be replaced with a length of fibers with W-style multi-clad refractive index profiles, i.e., ‘W-fibers’ (B. J. Ainslie et al.) or holey fibers (T. M. Monroe et al., ‘Holey Optical Fibers’ An Efficient Modal Model, J. Lightw. Techn., vol. 17, no. 6, pp. 1093-1102). Both W-fibers and holey fibers allow adjustable values of 2nd, 3rd and higher-order dispersion. Due to the small core size possible in W and holey fibers, larger values of 3rd order dispersion than in standard single-mode fibers can be obtained. The implementation is similar to the one shown in
An alternative PSM 2 with adjustable values of 2nd, 3rd and 4th order dispersion is shown in
The transfer loss of the whole fiber stretcher assembly is at least 25% due to the unavoidable 50% or greater loss incurred by transferring power from the LP11 mode to the LP01 mode without the use of optical mode-converters. Any residual energy in the LP01 mode in the 2nd stretcher fiber can be reflected with an optional reflective fiber grating 18 as shown in
The energy loss of the fiber stretcher assembly can be made to be insignificant by turning the 3rd stretcher fiber 17 into an Yb amplifier. This implementation is not separately shown.
When 4th-order dispersion is not significant, the 1st stretcher fiber 15 can be omitted. 4th order dispersion can also be compensated by using a 1st stretcher fiber with non-zero 3rd order dispersion, as long as the ratio of 3rd and 4th order dispersion is different between the 1st and 2nd stretcher fiber.
The Yb-doped fiber inside the AM13 can have an Yb doping level of 2.5 mole % and a length of 5 m. Both single-mode and multi-mode Yb-doped fiber can be used, where the core diameter of the fiber can vary between 1-50 μm; though the fundamental mode should be excited in case of a MM fiber to optimize the spatial quality of the output beam. Depending on the amount of required gain, different lengths of Yb-doped fiber can be used. To generate the highest possible pulse energies, Yb fiber lengths as short as 1 m can be implemented.
Pulse compression is performed in the PCM 4. The PCM 4 can contain conventional bulk optic components (such as the bulk diffraction grating pair shown in
Alternatively, a fiber or bulk Bragg grating can be used, or a chirped periodically poled crystal. The chirped periodically poled crystal combines the functions of pulse compression and frequency doubling (A. Galvanauskas, et al., ‘Use of chirped quasi-phase matched materials in chirped pulse amplification systems,’ U.S. application Ser. No. 08/822,967, the contents of which are hereby incorporated herein by reference) and operates in transmission providing for a uniquely compact system.
Other modifications and variations to the invention will be apparent to those skilled in the art from the foregoing disclosure and teachings.
In particular, the SM 1 can be used as a stand-alone unit to produce near bandwidth limited femtosecond pulses in the frequency range from 1.52-2.2 μm, and after frequency conversion in a nonlinear crystal also in the frequency range from 760 nm to 1.1 μm. The frequency range can be further extended by using a fluoride Raman-shifting fiber or other optical fibers with infrared absorption edges longer than silica. Using this technique wavelengths up to around 3-5 μm can be reached. In conjunction with frequency-doubling, continuous tuning from 760 nm to 5000 nm can be achieved. The pulse power in the 2 μm region can be further enhanced by using Tm or Ho-doped fiber. With such amplifiers, near bandwidth-limited Raman-soliton pulses with pulse energies exceeding 10 nJ can be reached in single-mode fibers in the 2 μm wavelength region. After frequency-doubling, femtosecond pulses with energies of several nJ can be obtained in the 1 μm region without the use of any dispersive pulse compressors. Such pulses can be used as high energy seed pulses for large-core multi-mode Yb amplifiers, which require higher seed pulse energies than single-mode Yb amplifiers to suppress amplified spontaneous emission.
An example of an ultra-wide tunable fiber source combining an Er-fiber laser pulse source 19 with a silica Raman-shifter 20, a Tm-doped amplifier 21 and a 2nd fluoride glass based Raman shifter 22 is shown in the SM 1c of
As yet another alternative for a SM, SM 1d is shown in
For optimum Raman self-frequency shift, the holey fiber dispersion should be optimized as a function of wavelength. The absolute value of the 3rd order dispersion of the holey fiber should be less than or equal to the absolute value of the 3rd order material dispersion of silica. This will help ensure that the absolute value of the 2nd order dispersion remains small over a substantial portion of the wavelength tuning range. Moreover the value of the 2nd order dispersion should be negative, and a 2nd order dispersion zero should be within 300 nm in wavelength to the seed input wavelength.
As yet another alternative for a seed source for an Yb amplifier, anti-Stokes generation in a length of anti-Stokes fiber can be used. After anti-Stokes generation, additional lengths of fiber amplifiers and Raman-shifters can be used to construct a widely wavelength-tunable source. A generic configuration is similar to the one shown in
As yet another alternative seed-source for an Yb amplifier, a passively modelocked Yb or Nd fiber laser can be used inside the SM. Preferably an Yb soliton oscillator operating in the negative dispersion regime can be used. To construct an Yb soliton oscillator, negative cavity dispersion can be introduced into the cavity by an appropriately chirped fiber grating 29, which is connected to output fiber 36 as shown in
An arrangement incorporating a holey fiber can be nearly identical to the system displayed in
Most straight-forward to implement, however, is an Yb oscillator operating in the positive dispersion regime, which does not require any special cavity components such as negative dispersion fiber Bragg gratings or holey fiber to control the cavity dispersion. In conjunction with a ‘parabolic’ Yb amplifier (or ordinary Yb amplifier), a very compact seed source for a high-power Yb amplifier system can be obtained. Such a Yb oscillator with an Yb amplifier 40 is shown in
The SM 1f in
To optimize the formation of parabolic pulses inside the Yb amplifier 40 inside the SM 1f, the input pulses should have a bandwidth small compared to the gain bandwidth of Yb; also the input pulse width to the Yb amplifier 40 should be small compared to the output pulse width and the gain of the Yb amplifier 40 should be as high as possible, i.e., larger than 10. Also, gain saturation inside the Yb amplifier 40 should be small.
As an example of a parabolic amplifier a Yb amplifier of 5 m in length can be used. Parabolic pulse formation is ensured by using a seed source with a pulse width of around 0.2-1 ps and a spectral bandwidth on the order of 3-8 nm. Parabolic pulse formation broadens the bandwidth of the seed source to around 20-30 nm inside the Yb amplifier 40, whereas the output pulses are broadened to around 2-3 ps. Since the chirp inside parabolic pulses is highly linear, after-compression pulse widths on the order of 100 fs can be obtained. Whereas standard ultrafast solid state amplifiers can tolerate a nonlinear phase shift from self-phase modulation only as large as pi (as well known in the state of the art), a parabolic pulse fiber amplifier can tolerate a nonlinear phase shift as large as 10*pi and higher. For simplicity, we thus refer to a large gain Yb amplifier as a parabolic amplifier. Parabolic amplifiers obey simple scaling laws and allow for the generation of parabolic pulses with spectral bandwidths as small as 1 nm or smaller by an appropriate increase of the amplifier length. For example, a parabolic pulse with a spectral bandwidth of around 2 nm can be generated using a parabolic amplifier length of around 100 m.
Since a parabolic pulse can tolerate large values of self-modulation and a large amount of spectral broadening without incurring any pulse break up, the peak power capability of a parabolic amplifier can be greatly enhanced compared to a standard amplifier. This may be explained as follows. The time dependent phase delay Φnl(t) incurred by self-phase modulation in an optical fiber of length L is proportional to peak power, i.e.
Φnl(t)=γP(t)L,
where P(t) is the time dependent peak power inside the optical pulse. The frequency modulation is given by the derivative of the phase modulation, i.e., δω=γL[∂P(t)/∂t]. For a pulse with a parabolic pulse profile P(t)=P0[1−(t/t0)2], where (−t0<t<t0), the frequency modulation is linear. It may then be shown that indeed the pulse profile also stays parabolic, thus allowing the propagation of large peak powers with only a resultant linear frequency modulation and the generation of a linear pulse chirp.
The chirped pulses generated with the Yb amplifier 40 can be compressed using a diffraction grating compressor as shown in
In addition to the passively modelocked Yb fiber laser 44 shown in
In
In its simplest configuration, the FDM 45 consists of a length of optical fiber 46 (the delivery fiber). For a parabolic amplifier, the delivery fiber 46 can be directly spliced to the Yb amplifier 7 without incurring any loss in pulse quality. Rather, due to the parabolic pulse profile, even for large amounts of self-phase modulation, an approximately linear chirp is added to the pulse allowing for further pulse compression with the PCM 4. The PCM 4 can be integrated with the FDM 45 by using a small-size version of the bulk diffraction grating compressor 14 shown in
To obtain pulse widths shorter than 50 fs, the control of third order and higher-order dispersion in a FDM module or in an optional PSM becomes significant. The control of higher-order dispersion with a PSM was already discussed with reference to
Alternative embodiments of PSMs are shown in
Alternatively, the pulses can be left chirped and compensated with the compressor after the regenerative amplifier. This would mean not utilizing the PCM. This design would place additional design challenges on the dispersion correction in the PSM.
To avoid the use of W-fibers or the LP11 mode in the PSM, an alternative embodiment of a PSM as shown in
As yet another alternative embodiment of a PSM for the compensation of higher-order dispersion the arrangement in
To increase the amplified pulse energy from an Yb amplifier to higher pulse energies, pulse picking elements and further amplification stages can be implemented as shown in
Amplifier module AM13a can be designed as a parabolic amplifier producing pulses with a parabolic spectrum. Equally, the parabolic pulses from AM13a can be transformed into pulses with a parabolic pulse spectrum in a subsequent length of pulse-shaping or pulse stretching fiber 53 as also shown in
Once the pulses are stretched, the detrimental influence of self-phase modulation in subsequent amplifiers can be minimized by using flat-top pulse shapes. A flat-top pulse shape can be produced by inserting an optional amplitude filter 54 as shown in
An amplitude filter as shown in
where P0 is the peak power of the pulse and S(ω) is the normalized pulse spectrum. Leff is the effective nonlinear length Leff=[exp(gL)−1]/g, where L is the amplifier length and g is the amplifier gain per unit length. Thus by accurately controlling the spectrum of strongly chirped pulses with an amplitude filter as shown in
As another embodiment for the combination of an amplifier module with a pulse picker, the configuration displayed in
An exemplary integrated fiber polarizer in accordance with the invention comprises a polarization-maintaining undoped polarizer fiber (PF), with two orthogonal polarization axes, where the loss along one polarization axis is significantly higher than the loss along the other polarization axis. Alternatively, a very short section (less than 1 cm) of non-birefringent fiber (i.e., non-polarization-maintaining fiber) can be sandwiched between two sections of polarization-maintaining fiber, where the polarization axes of the polarization-maintaining fibers are aligned with respect to each other. By side-polishing the non-birefringent fiber, e.g., down to the evanescent field of the fiber core, along one of the axes of the birefringent fiber, and coating the polished region with metal, high extinction polarization action can be obtained along one of the axes of the birefringent fiber. The design of side-polished fiber polarizers is well known in the field and not discussed further here.
For optimum laser operation, the fiber polarization axes of the PF are aligned parallel to the polarization axes of the gain fiber 101. To ensure stable modelocked operation, the polarizer preferably effectively eliminates satellite pulses generated by any misalignment between the polarization axes of the PF and the gain fiber 101.
Neglecting any depolarization in the all-fiber polarizer itself, it can be shown by applying a Jones matrix calculation method that for a misalignment of the polarization axes of gain fiber 101 and fiber polarizer 105 by α degrees, the linear reflectivity R from the right-hand side of the cavity varies approximately between R=1−0.5 sin2 2α and R=1 depending on the linear phase in the gain fiber 101. If the group delay along the two polarization axes of the gain fiber is larger than the intra-cavity pulse width, any satellite pulse is suppressed by sin4 α after transmission through the polarizer. Typical fiber splicing machines can align polarization-maintaining fibers with an angular accuracy of less than ±2°; hence any reflectivity variation due to drifts in the linear phase between the two polarization eigenmodes of fiber 101 can be kept down to less than 3×10−3, whereas (for sufficiently long fibers) any satellite pulses obtained after transmission through the polarizer can be kept down to less than 6×10−6 when using an integrated polarizer.
The chirped fiber Bragg grating 106 is preferably spliced to the PF 105 at splice position 108 and written in non-polarization-maintaining fiber. In order to avoid depolarization in the fiber Bragg grating, the Bragg grating pig-tails are preferably kept very short, e.g., a length smaller than 2.5 cm is preferable between splice locations 108 and 109. To obtain a linear polarization output, a polarization-maintaining fiber pig-tail is spliced to the left-side of the fiber Bragg grating at splice location 109. The laser output is obtained at a first fiber (or cavity) end 111, which is preferably angle-cleaved to avoid back-reflections into the cavity. An alternative preferred design is with the fiber grating written in polarization-maintaining fiber.
Fiber Bragg grating 106 serves two functions. First, it is used as an output mirror (i.e., it feeds part of the signal back to the cavity) and, second, it controls the amount of cavity dispersion. In the present implementation, the chirped fiber Bragg grating has a negative (soliton-supporting) dispersion at the emission wavelength in the wavelength region near 1060 nm and it counter-balances the positive material dispersion of the intra-cavity fiber. To produce the shortest possible pulses (with an optical bandwidth comparable to or larger than the bandwidth of the gain medium), the absolute value of the grating dispersion is selected to be within the range of 0.5-10 times the absolute value of the intra-cavity fiber dispersion. Moreover, the fiber Bragg grating is apodized in order to minimize any ripple in the reflection spectrum of the grating. Accordingly, the oscillation of chirped pulses is enabled in the cavity, minimizing the nonlinearity of the cavity and maximizing the pulse energy. Chirped pulses are characterized in having a pulse width which is longer than the pulse width that corresponds to the bandwidth limit of the corresponding pulse spectrum. For example the pulse width can be 50%, 100%, 200% or more than 1000% longer than the bandwidth limit.
Alternatively, the oscillation of chirped pulses is also enabled by using negative dispersion fiber in conjunction with positive dispersion chirped fiber Bragg gratings. Pulses with optical bandwidth comparable to the bandwidth of the gain medium can also be obtained with this alternative design.
A SAM 112 at a second distal fiber end 113 completes the cavity. In an exemplary implementation a thermally expanded core (TEC) 110 is implemented at cavity end 113 to optimize the modelocking performance and to allow close coupling of the SAM 112 to the second fiber end 113 with large longitudinal alignment tolerances. Etalon formation between the fiber end 113 and the SAM 112 is prevented by an anti-reflection coating deposited on fiber end 113 (not separately shown). In the vicinity of the second fiber end 113, fiber 101 is further inserted into ferrule 114 and brought into close contact with SAM 112. Fiber 101 is subsequently fixed to ferrule 114 using, for example, epoxy and the ferrule itself is also glued to the SAM 112.
The pump laser 104 is coupled into the gain fiber 101 via a lens system comprising, for example, two lenses 115 and 116 and a V-groove 117 cut into fiber 101. Such side-coupling arrangements are described in, for example, U.S. Pat. No. 5,854,865 ('865) to L. Goldberg et al. Alternatively, fiber couplers can be used for pump light coupling.
An exemplary design for a SAM in accordance with the present invention is shown in
In order to cover a broad spectral range (e.g., greater than 100 nm) metallic mirrors are preferred. When using a metallic mirror it is advantageous to remove the substrate (InP) by means of etching. When using HCl as an etching solvent the etching selectivity between InGaAsP and InP can be low, depending on the compound composition of InGaAsP. An etch-stop layer is beneficial between the substrate and the InGaAsP layer. InGaAs can be a proper etch-stop layer. When adding an InGaAs layer with a band-gap wavelength shorter than 1.03 μm, lattice relaxations can be avoided by keeping the thickness below 10 nm.
The InGaAsP layer can further be anti-reflection coated with layer 204 on its upper surface to optimize the performance of the SAM. Because of the saturable absorption by InGaAsP, the reflectivity of the SAM increases as a function of light intensity, which in turn favors the growth of short pulses inside the laser cavity. The absence of Al in the saturable absorber layer prevents oxidization of the semiconductor surfaces in ambient air and thus maximizes the life-time and power handling capability of the structure.
Instead of InGaAsP, any other Al-free saturable semiconductor can also be used in the construction of the SAM. Alternatively, Al-containing semiconductors can be used in the SAM with appropriately passivated surface areas. Surface passivation can, for example, be accomplished by sulfidization of the semiconductor surface, encapsulating it with an appropriate dielectric or with an Al-free semiconductor cap layer. An AlGaInAs absorber layer grown lattice-matched on InP can be surface-passivated with a thin (about 10 nm range) cap layer of InP. AlGaInAs with a higher band gap energy than the absorber layer can also be used for a semiconductor Bragg reflector in combination with InP. Among concepts for semiconductor Bragg mirrors lattice-matched to InP, an AlGaInAs/InP combination has an advantage over an InGaAsP/InP Bragg reflector due to its high refractive index contrast.
Instead of a bulk semiconductor saturable absorber, a MQW saturable absorber structure as shown in
The control of the response time of the saturable absorption for concomitant existence of fast and slow time constants is realized by introducing carrier trap centers with depth controlled H+ (or other ions) implantation. The implantation energy and dose are adjusted such that part of the absorbing semiconductor film contains a minimal number of trap centers. For example the semiconductor layer with the minimal number of trap centers can be selected to be at the edge of the optical penetration range of exciting laser radiation. Such a design serves only as an example and alternatively any semiconductor area within the optical penetration range can be selected to contain a minimal number of trap centers. Hence distinctive bi-temporal carrier relaxation is obtained in the presence of optical excitation. As an illustration of depth selective ion implantation,
The obtained bi-temporal carrier life-time obtained with the semiconductor film with a proton concentration as shown in
When employing this absorber in the laser system described with respect to
We emphasize that the description for
The preferred implantation parameters for H+ ions in GaAs or InP related materials including MQW absorbers are as follows: The doses and the implantation energies can be selected from 1012 cm−2 to 1017 cm−2 and from 5 keV to 200 keV, respectively, for an optically absorbing layer thickness between 50 nm and 2000 nm. For MQW absorbers, the selective ion-implantation depth is rather difficult to measure because the shallow MQW falls into the implantation peak in
a illustrates an alternative implementation of the fiber end and SAM coupling in
Instead of TEC cores, more conventional lenses or graded index lenses can be incorporated between the fiber end and the SAM to optimize the beam diameter on the SAM. Generally, two lenses are required. A first lens collimates the beam emerging from the fiber end, and a second lens focuses the beam onto the SAM. According to present technology, even conventional lenses allow the construction of a very compact package for the second fiber end. An implementation with two separate collimation and focusing lenses is not separately shown. To minimize unwanted back reflections into the fiber cavity and to minimize the number of components, a single lens can be directly fused to the fiber end as depicted in
To minimize aberrations in assembly 306, an additional lens can also be incorporated between lens 307 and SAM 303. Such an assembly is not separately shown. Alternatively, a lens can be directly polished onto fiber 304; however, such an arrangement has the disadvantage that it only allows a beam size on the SAM which is smaller than the beam size inside the optical fiber, thereby somewhat restricting the design parameters of the laser. To circumvent this problem, a lens surface can be directly polished onto the surface of a TEC; such an implementation is not separately shown. Another alternative is to exploit a graded-index lens design attached directly onto the fiber tip to vary the beam size on the SAM. In the presence of air-gaps inside the oscillator a bandpass filter 308 can be incorporated into the cavity, allowing for wavelength tuning by angular rotation as shown, for example, in
Passive modelocking of laser cavity 100 (
When pumping with an optical power of 1.0 W at a wavelength of 910 nm, the laser produced short chirped optical pulses with a full width half maximum width of 1.5 ps at a repetition rate of 50 MHz. The average output power was as high as 10 mW. The pulse bandwidth was around 2 nm and hence the pulses were more than two times longer than the bandwidth-limit which corresponds to around 800 fs.
Alternatively, a fiber grating 106 with a dispersion of −0.1 ps2, closely matching the dispersion of the intra-cavity fiber, was implemented. The fiber grating had a reflectivity of 9% and a spectral bandwidth of 22 nm centered at 1050 nm. The grating was manufactured with a phase mask with a chirp rate of 320 nm/cm. The laser then produced chirped optical pulses with a full-width half maximum width of 1.0 ps at a repetition rate of 50 MHz with an average power of 25 mW. The pulse spectral bandwidth was around 20 nm and thus the pulses were around 10 times longer than the bandwidth limit, which corresponds to around 100 fs. The generation of pulses with a pulse width corresponding to the bandwidth limit was enabled by the insertion of a pulse compressing element; such elements are well known in the state of the art and are not further discussed here. The generation of even shorter pulses can be generated with fiber gratings with a bandwidth of 40 nm (and more) corresponding to (or exceeding) the spectral gain bandwidth of Yb fibers.
Shorter pulses or pulses with a larger bandwidth can be conveniently obtained by coupling the fiber output into another length of nonlinear fiber as shown in
In addition to cladding pumped fiber lasers, core-pumped fiber lasers can be constructed in an integrated fashion. Such an assembly is shown in
In addition to chirped fiber gratings, unchirped fiber gratings can also be used as output couplers. Such cavity designs are particularly interesting for the construction of compact Er fiber lasers. Cavity designs as discussed with respect to
In the absence of a fiber grating, a mirror can be deposited or attached to one end of the fiber cavity. The corresponding cavity design 600 is shown in
A second fiber polarizer 608 is spliced between WDM 605 and polarization-maintaining output coupler 609 to minimize the formation of satellite pulses, which can occur when splicing sections of polarization maintaining fiber together without perfect alignment of their respective polarization axes, as discussed in U.S. patent application Ser. No. 09/809,248. Typically, coupler 609 has a coupling ratio of 90/10 to 50/50, i.e., coupler 609 couples about 90-50% of the intra-cavity signal out to fiber pig-tail 610. Pig-tail 610 can be spliced to a fiber isolator or additional fiber amplifiers to increase the pulse power. The second cavity end is terminated by mirror 611. Mirror 611 can be directly coated onto the fiber end face or, alternatively, mirror 611 can be butt-coupled to the adjacent fiber end.
The increase in stability of cavity 600 compared to a cavity where the output coupler fiber, the WDM fiber and gain fiber 601 are directly concatenated without intra-fiber polarizing stages, can be calculated using a Jones matrix formalism even when coherent interaction between the polarization axes of each fiber section occurs.
Briefly, due to the environmental sensitivity of the phase delay between the polarization eigenmodes of each fiber section, for N directly concatenated polarization-maintaining fibers the reflectivity of a fiber Fabry-Perot cavity can vary between R=1 and R=1−(N×α)2, where α is the angular misalignment between each fiber section. Further, it is assumed that α is small (i.e., α<<10°) and identical between each pair of fiber sections. Also, any cavity losses are neglected. In fact, it is advantageous to analyze the possible leakage L into the unwanted polarization state at the output of the fiber cavity. L is simply given by L=1−R. For the case of N concatenated fiber sections, the maximum leakage is thus (N×α)2.
In contrast, a cavity containing N−1 polarizers in-between N sections of polarization-maintaining fiber is more stable, and the maximum leakage is L=2×(N−1)α2. Here, any depolarization in the fiber polarizers itself is neglected. For instance, in a case where N=3, as in cavity 600, the leakage L into the wrong polarization axis is 2×(3−1)/33=4/9 times smaller compared to a cavity with three directly concatenated fiber sections. This increase in stability is very important in manufacturing yield as well as in more reproducible modelocked operation in general.
In constructing a stable laser, it is also important to consider the construction of WDM 605 as well as output coupler 609. Various vendors offer different implementations. An adequate optical representation of such general polarization-maintaining fiber elements is shown in
In order to ensure pulse stability inside a passively modelocked laser, the group-velocity walk-off along the two polarization axes of fiber sections 701, 702 should then be longer than the full-width half maximum (FWHM) pulse width of the pulses generated in the cavity. For example, assuming a birefringent fiber operating at a wavelength of 1550 nm with a birefringence of 3×104 corresponding to a polarization beat length of 5 mm at 1550 nm, the stable oscillation of soliton pulses with a FWHM width of 300 fs requires pig-tails with a length greater than 29 cm. For 500 fs pulses, the pig-tail length should be increased to around 50 cm.
Referring back to
Alternatively, a fiber-pig-tail can be butt-coupled to mirror 611 and also be used as an output fiber pigtail. Such an implementation is shown in
Similarly, a cladding pumped version of cavity 600 can be constructed. Cavity 900 shown in
The cavity designs discussed with respect to
a shows a representative modelocked Fabry-Perot fiber laser cavity 1000, producing a linear polarization state oscillating inside the cavity containing one (or more) sections of non-polarization maintaining fiber 1001 and one (or more) sections of polarization maintaining fiber 1002, where the length of fiber section 1001 is sufficiently short so as not to degrade the linear polarization state inside the fiber laser cavity, more generally a predominantly linear polarization state is oscillating everywhere within the intracavity fiber. The fiber laser output can be obtained from cavity end mirrors 1003 or 1004 on either side of the cavity. To suppress the oscillation of one over the other linear polarization state inside the cavity, either fiber 1001 or 1002 has a polarization dependent loss at the emission wavelength.
b shows a representative modelocked Fabry-Perot fiber laser cavity 1005, producing a linear polarization state oscillating inside the cavity containing two (or more) sections of polarization maintaining fibers 1006, 1007, where the length of fiber sections 1006, 1007 is sufficiently long so as to prevent coherent interaction of short optical pulses oscillating inside the cavity and propagating along the birefringent axes of fibers 1006, 1007. Specifically, for an oscillating pulse with a FWHM width of τ, the group delay of the oscillating pulses along the two polarization axes of each fiber should be larger than τ. For oscillating chirped pulses τ represents the bandwidth-limited pulse width that corresponds to the oscillating pulse spectrum. Cavity 1005 also contains end mirrors 1008 and 1009 and can further contain sufficiently short sections of non-polarization maintaining fiber as discussed with reference to
c shows a representative modelocked Fabry-Perot fiber laser cavity 1010, producing a linear polarization state oscillating inside the cavity containing one (or more) sections of polarization maintaining fiber 1011, 1012 and one (or more) sections of polarizing fiber (or all-fiber polarizer) 1013, where the length of fiber sections 1011, 1013 is not sufficient to prevent coherent interaction of short optical pulses oscillating inside the cavity and propagating along the birefringent axes of fibers 1011, 1013, where the polarizing fiber is sandwiched between the sections of short polarization maintaining fiber. Cavity 1010 further contains cavity end mirror 1014 and 1015 and can further contain short sections of non-polarization maintaining fiber as discussed with reference to
To enable the oscillation of short pulses with a bandwidth comparable to the bandwidth of the gain medium 1402, saturable absorber mirror 1403 contains a bi-temporal saturable absorber, constructed with a bi-temporal life-time comprising a 1st short life-time of <5 ps and a 2nd long life-time of >50 ps. More preferable is a first life-time of <1 ps, to allow pulse shaping of pulses as short as 100 fs and shorter. By selecting the penetration depth of the implanted ions into the saturable absorber, even tri-temporal saturable absorbers can be constructed.
The wide-bandwidth grating is preferably selected to approximately match the dispersion of the intra-cavity fibers. The wide-bandwidth grating can be made in short non-polarization maintaining fibers and it can be made also in polarization maintaining fibers. In order to suppress detrimental effects from cross coupling between the two polarization axes of the fiber grating, coupling to cladding modes in such large bandwidth fiber gratings should be suppressed. Gratings with suppressed coupling to cladding modes can be made in optical fibers with photosensitive core and cladding area, where the photosensitive cladding area is index-matched to the rest of the cladding. Such fiber designs are well known in the state of the art and can for example be manufactured with an appropriate selection of germania and fluorine doping in the core and cladding regions and such fiber designs are not further discussed here. Because of the large generated bandwidth, splicing of such polarization maintaining gratings to the rest of the cavity without coherent coupling between the linear polarization eigenmodes is no problem. Alternatively, the fiber gratings can be written directly into the photosensitive gain fiber, with an index and dopant profile that suppresses coupling to cladding modes in the fiber grating.
To sustain large spectral bandwidth, fiber grating 1404 has preferably a spectral bandwidth >20 nm. A splice 1407 (or an equivalent bulk optic lens assembly) is used to connect the output of fiber laser 1401 to nonlinear fiber 1408 to be used for additional spectral broadening of the output of the fiber laser. For example fiber 1408 can comprise a highly nonlinear dispersion-flattened holy fiber. In conjunction with such fiber, smooth broad-bandwidth spectral profiles with bandwidths exceeding 100 nm can be generated. These spectral outputs can be used directly in high precision optical coherence tomography.
The pulses at the output of fiber 1408 are generally chirped and a dispersion compensation module 1409 can be inserted after the output from fiber 1408 for additional pulse compression. The dispersion compensation module can be spliced directly to fiber end 1408 when optical fiber is used for dispersion compensation. Alternatively, the dispersion compensation module can comprise two (or one) bulk grating (or prism) pair(s). Such bulk optic elements for dispersion compensation are well known in the state of the art and are not further discussed here. Coupling into and out of a bulk dispersion compensating module is obtained via lenses 1410 and 1411. The output can also be from the other end of the cavity. The pulses generated after pulse compression can be as short as 20-200 fs. As mentioned previously this pulse compression module is optional and the dispersion compensation needed for this oscillator can be compensated by the pulse stretcher before and pulse compressor after the regenerative amplifier.
A fiber amplifier 1412 can also be added if further pulse energy is necessary.
Note that the discussion with respect to
Though the discussion of the laser system with respect to
An embodiment with the fewest bulk optic components in the optical path is shown in
The embodiment in
As shown in
The exemplary laser system 550 is shown to further comprise a second monitor component 574 that monitors a performance parameter of the system after the amplifier 560. The monitor 574 may similarly comprise a sensor and controller. The monitor 574 can then issue adjustment commands to a second driver 576 that implements those adjustment commands at the amplifier 560.
The monitoring of the system performed by the exemplary monitors 570 and/or 574 may comprise for example an optical detector and electronics that monitors optical intensity or power or other relevant parameter such as, e.g., frequency and spectrum. In response to such measurement, the monitor and the driver may induce changes in the oscillator and/or the amplifier by for example adjusting the pump intensity and/or rate, or adjusting the operating temperature. Temperature control of the oscillator can stabilize the gain dynamics as well as frequency fluctuations. Temperature control of the amplifier can also be used to stabilize the gain dynamics.
Other configurations for providing feedback to control the operation of the laser system may also be employed. For example, more or less feedback loops may be included. The electronics associated with these feedback loops are further described in Ser. No. 10/813,173 (incorporated by reference herein). A particularly important electronic control is to control the gain of the fiber amplifier. At 1 KHz repetition rate and lower, the gain of the fiber amplifier could be reduced between pulses to conserve the lifetime of the laser diode. Also the gain needs to be reduced on the fiber amplifier if a signal is lost from the short pulse source to protect from optical damage to the fiber amplifier or subsequent optical elements. The loops may involve electronics that perform operations such as calculations to determine suitable adjustments to be introduced. Examples are the mode-lock start-up and search algorithms that are disclosed in Attorney Docket No. A8828 (incorporated by reference herein). The start-up algorithm is shown in
A good Polarization Extinction Ratio (PER) is an important factor in maintaining good temporal pulse quality in a fiber-based ultrafast source for a regenerative amplifier. Poor polarization extinction creates ripple on the spectrum and on the chirped pulse. In various preferred embodiments, the light in the laser is linearly polarized. The degree of the linear polarization may be expressed by the polarization extinction ratio (PER), which corresponds to a measure of the maximum intensity ratio between two orthogonal polarization component. In certain embodiments, the polarization state of the source light may be maintained by using polarization-maintaining single-mode fiber. For example, the pigtail of the individual modular device may be fabricated with a polarization-maintaining fiber pigtail. In such cases, the PER of each modular stage may be higher than about 23 dB. Ensuring a high polarization extinction ratio throughout a series of modules is challenging despite the use of single mode polarization maintaining fiber. Degradation of the PER can occur at the fiber ferrule, fiber holder, or fusion splice in the series of modules.
Levels of PER above 23 dB may be obtained in a system by utilizing linear-polarizing optical components in the modules. Use of linear-polarizing components in the modules within systems that contain polarization degrading elements such as a fiber ferrule, fiber holder, or fusion splice is advantageous. The linear polarizers counter the superposition of the phase shift from each polarization degrading element. A superposed phase shift of 10 degrees may reduce the PER to about 15 dB in which case intensity fluctuation through a linear polarizer might be more than about 4%. In contrast, by embedding linear polarizers throughout the series of modules, the PER of the aggregate system can be substantially controlled such that the intensity fluctuation is below about 1%, provided that the PER of the individual module and splice is above about 20 dB.
a illustrates one embodiment of a module that can be utilized for polarization correction or as variable attenuation. It is a variable attenuator module 730 comprising a housing 732 that contains optical components for providing a controllable amount of optical attenuation. The housing 732 may be sealed and thermally insulated as well. A first optical fiber connector 734 comprising an optical fiber 736 having an angle polished or cleaved end face passes through one sidewall of the housing 732 into an inner region of the housing containing the plurality of optical components. These optical components include a first lens 738 for collecting and preferably collimating light output from the optical fiber 736, a variable wave plate 740 and a polarization selective optical element 742. A second optical fiber connector 744 comprising an optical fiber 745 having an angle polished or cleaved end face passes through another sidewall of the housing 732 into the inner region containing the optical components. The variable waveplate 740 comprises a rotatable waveplate mounted on a rotatable wheel 746 and the polarization selective optical element 742 comprises a polarization beamsplitter such as a MacNeille prism. A second lens 748 disposed between the polarization selective optical element 742 couples light between the polarization beamsplitter 742 and the second optical fiber 745. An optical path is formed from the first optical fiber 736 through the waveplate 740 and prism 742 to the second optical fiber connector 744.
The waveplate 740 can be rotated to vary the distribution of light into orthogonal polarizations. The polarization beamsplitter 742 can be used to direct a portion of the light out of the optical path between the first and second fiber connectors 734, 744, depending on the state of the waveplate 740. Accordingly, a user, by rotating the waveplate 740 and altering the polarization of light can control the amount of light coupled between the first and second optical fiber connectors 734, 744 and thereby adjust the level of attenuation.
Preferably, the optical elements such as the first and second lenses 738, 748, the rotatable waveplate 740 and the MacNeille polarizer 742 comprise micro-optics or are sufficiently small to provide for a compact module. The elements in the housing 732 may be laser welded or otherwise securely fastened to a base of the housing. The housing 732 may be sealed and thermally insulated as well. In various preferred embodiments, these modules conform to Telcordia standards and specifications.
A particularly preferred embodiment for a fiber solid-state regenerative amplifier system (2000) is shown in
In this next embodiment an alternative source of the ultrafast pulses is a laser-diode or microchip laser. This embodiment is shown in
Since there is no interplay between dispersion and self-phase modulation in this design, the pulse width stays the same as the original pulse width. The nearly linear chirp is created by the shape of the pulse. Such a fiber is normally called a “compression fiber”. We propose to replace this “compression fiber” with an amplifier fiber. The output of the amplifier will be a chirped pulse that can be compressed in a compressor. This saves the need of a stretcher.
For pulse energies significantly greater than 1 microjoule, the single mode beam should be further amplified in a multimode fiber. This chirped pulse source is ideal for amplification of ultrashort pulses by chirped pulse amplification in a regenerative amplifier. The pulse is then compressed after amplification. In this case the microchip 71 was operated at 0.5 μJ, and produced 250 ps, pulses and operating at the repetition rate of the regenerative amplifier. The compression fiber 62 is now a multimode amplifier fiber that amplified a single mode with a mode-field diameter of 17 μm. The pulse was then amplified to 30 microjoules where Raman limited the amplification. This pulse is now a chirped 250 ps pulse. It is further amplified in a solid state regenerative amplifier and compressed in a bulk grating compressor to typically less than 1 ps.
After the bandpass filter 119, an output power of 5 milliwatts was obtained and a fiber stretcher 120 was used to stretch the pulses to a width of approximately 100 picoseconds. The fiber stretcher employed for producing stretched pulses had a length of approximately 200 meters and was based on conventional polarization maintaining single-mode step-index fiber. In
A subsequent Yb-based polarization maintaining pre-amplifier 121 amplifies the stretched pulses to an average power of 500 milliwatts. A pulse picker 122, based on an acousto-optic modulator and pig-tailed with polarization maintaining fiber, reduces the repetition rate of the pulses to 200 kilohertz, resulting in an average power of 1 milliwatt. The pulses from the pulse picker 122 were subsequently injected into a large-mode polarization maintaining Yb fiber power amplifier 123 and amplified to an average power of 950 milliwatts. The Yb power amplifier had a length of 3 meters and the fundamental mode spot size in the Yb power amplifier was around 25 micrometers. All fibers were either spliced together with their polarization axes aligned or connected to each other (with their polarization axes aligned) with appropriate mode-matching optics (not shown). The power amplifier 123 was cladding pumped via a lens 124 with a pump source 125, delivering a pump power of about 10 watts at a wavelength of 980 nanometers. A beam splitting mirror 126 was implemented to separate the pump light from the amplified signal light. The amplified and stretched pulses from the power amplifier 123 are further amplified in a bulk solid state regenerative amplifier 129. The output pulses from the regenerative amplifier 129 were compressed in a conventional bulk optics compressor 127 based on a single diffraction grating with a groove density of 1200 lines/mm, operating near the Littrow angle. Such bulk optics compressors are well known in the state of the art and are not further explained here. After the bulk optics compressor 127, the output 128 will contain pulses with a full-width half-maximum (FWHM) width of around 330 femtoseconds and pulse energies around 1 millijoule. Alternative designs should be feasible including a system without the power amplifier. However, in this case the power amplifier is operating as the nonlinear fiber amplifier that is able to correct for higher order dispersion mismatch between the fiber stretcher and the bulk compressor.
Because stretched pulses can accumulate significant levels of third-order dispersion in the presence of self-phase modulation, gain-narrowing, gain-pulling and gain depletion, we refer to such pulses as cubicons. More generally, we can define a cubicon as a pulse that produces controllable levels of at least linear and quadratic pulse chirp in the presence of at least substantial levels of self-phase modulation (corresponding to a nonlinear phase delay >1) that can be at least partially compensated by dispersive delay lines that produce significant levels of second and third-order dispersion as well as higher-order dispersion. (Please note that for the compensation of linear pulse chirp, a dispersive delay line with second order dispersion is required, whereas for the compensation of quadratic pulse chirp, a dispersive delay line with third order dispersion is required and so on for higher orders of pulse chirp.) For a dispersive delay line to produce a significant level of 2nd and 3rd as well as possibly higher-order dispersion, the stretched pulses are typically compressed by more than a factor of 30. In addition cubicons can also be formed in the presence of resonant amplifier dispersion, gain narrowing, gain pulling as well as gain depletion, where we refer to gain depletion as an appreciable reduction in gain due to a single pulse. If a high power mode-locked oscillator an undoped fiber can be utilized to create the self-phase modulation. Spectral filtering will most likely be necessary to obtain the appropriate pulse shape to the chirped pulse. The chirped pulse width will need to be further expanded before amplification in the regenerative amplifier.
The importance of the pulse picker 122 has been described in Ser. No. 10/960,923 in that it alleviates the specifications on the optical switch in the regenerative amplifier. A further advantage is that it can be utilized as a variable attenuator for controlling the buildup time in the regenerative amplifier. An AO switch can be used here, however EO switches and EA switches are available in modules that conform to Telcordia standards and specifications. As pointed out in Ser. Nos. 10/437,057 and 10/606,829, it often takes two switches since the standard on off discrimination is 30 db while for lowering the rep rate from 30 MHz to 1 KHz requires an on off discrimination of more than 50 db for the majority of the energy to be in the one pulse operating at the lower repetition rate. Another use of the pulse picker is as a variable attenuator to control the nonlinearities in the fiber for dispersion correction. In cubicon amplification the nonlinearities are critical for dispersion control and the variable attenuation feature of the pulse pickers is a means for controlling the nonlinear affects in the fibers. Other variable attenuators can be used such as described in Ser. No. 10/814,319. Other means of controlling the nonlinearities of the fiber amplifier are utilizing the control of the fiber amplifier output as described above. These include varying the gain or temperature of the fiber amplifier by measuring the spectrum and or the output intensity from the fiber amplifier. Controlling the spectrum and the intensity accurately for cubicon amplification can be implemented.
The embodiment of a short pulse source in the picosecond and nanosecond range amplified in a fiber amplifier and amplified in a bulk amplifier is disclosed in application Ser. No. 10/927,374 (incorporated by reference herein) This system in some cases will have better performance when the bulk amplifier is utilized as a regenerative amplifier. This embodiment is shown in
Bulk solid-state regenerative amplifiers are also useful to increase the energy of pulses generated with fiber based chirped pulse amplification systems. Chirped pulse amplification is generally employed to reduce nonlinearities in optical amplifiers. The implementation of chirped pulse amplification is most useful for the generation of pulses with a width <50 ps. Due to the limited amount of pulse stretching and compression that can be achieved with chirped pulse amplification schemes, stretched pulses with an initial pulse width exceeding 1-5 ns are generally not implemented. Hence optical damage limits the achievable pulse energies from state of the art fiber based chirped pulse amplification systems (assuming fiber power amplifiers with a core diameter of 30 μm) to around 1 mJ. Single stage bulk solid state amplifiers can increase the achievable pulse energies normally by a factor of 10 while a regenerative amplifier has a gain of 106. Therefore a regenerative amplifier can be preferable and give flexibility at a cost of complexity. One advantage is significantly lower pulse energies can be utilized from the fiber amplifier. A generic scheme 500 for the amplification of the output of a fiber based chirped pulse amplification system in a bulk optical amplifier is shown in
In an exemplary embodiment, fiber oscillator 501 generates 5 ps pulses, which are stretched by a chirped fiber grating stretcher to a width of 1 ns. After amplification in the fiber amplifier chain a pulse energy of 50 μJ is obtained at a repetition rate of 10 kHz. Further amplification in a Nd:YVO4 solid state booster amplifier generates a pulse energy of 2 mJ. After recompression in the bulk grating compressor 10 ps pulses with an energy of 1 mJ are obtained. To ensure a compact design for the bulk grating compressor, preferably grisms with a groove density of 2800 l/mm are implemented. The whole compressor can then fit into an area of about 0.6×0.2 m by folding the optical beam path only once.
As discussed above, a burst of multiple pulses with different wavelengths, different pulse widths and different temporal delays may be desired. Referring to
If the distances between the compression and stretcher gratings are not equalized as described above, then multiple pulses will appear at the output. If the distances are not equal between the different sections than the temporal delays will not be equal. This can be beneficial for applications such as micro-machining. By varying the stretching and compression ratios, pulses with different pulse widths can be generated. A single broadband compression grating can be used when generating multiple pulses.
The utilization of the regenerative amplifier is not as flexible as an all fiber amplifier system for modification of the pulse shape. For example, long pulse widths are limited to repetitive features equal to the round trip time of the regenerative amplifier, e.g., approximately 10 nanoseconds. For a regenerative amplifier, the pulse train created by the gratings needs to be less than the round trip time of the regenerative amplifier.
Another embodiment of a multiple pulse source is shown in
Another embodiment of this is to utilize laser diodes at different wavelengths or polarization states and then combine these optically either with wavelength fiber combiners such as the wavelength router utilized in multiple wavelength telecomm systems or by fiber splitters as shown in
1. c(2LN+L1+L4)
2. c(2LN+L1+L3)
3. c(2LN+L2+L4)
4. c(2LN+L2+L3)
| Number | Name | Date | Kind |
|---|---|---|---|
| 20050111500 | Harter et al. | May 2005 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20060120418 A1 | Jun 2006 | US |
